Conventionally, the substrates employed in semiconductor devices are in the form of ingots manufactured from compound semiconductor crystals, for example. Examples of widely employed methods of manufacturing ingots are the horizontal Bridgman method, Czochralski method, sublimation method, and zone melting method. The ingot obtained is generally cut into plates, externally processed, surface ground, and employed as a semiconductor wafer. Further, when manufacturing high-performance semiconductor devices, a film with few crystal defects and of low impurity concentration is epitaxially grown on a compound semiconductor substrate that has been processed from an ingot.
Further, it is sometimes difficult to manufacture a compound semiconductor as an ingot, and it is sometimes difficult to obtain large-area ingots for technical and cost reasons. In such cases, the means of forming an epitaxial layer of a desired compound semiconductor crystal on a relatively readily available semiconductor wafer in the form of an Si substrate or the like (heteroepitaxy) is sometimes employed.
However, in heteroepitaxy, there is mismatching of the lattice constants of the substrate crystal and the crystal grown on the layer above it. Thus, crystal defects resulting from this mismatch develop at the interface of the substrate crystal and the crystal grown over it. In particular, in compound semiconductor crystals, a greater variety of crystal defects develop than in crystals comprised of a single element, such as Planar defects (Anti-Phase boundaries) developing at boundaries between regions where the stacking order of constituent elements is reversed and defects (twins) developing where the crystal is partially rotated. Such crystal defects must be reduced to the extent possible in the manufacturing of semiconductor devices because, for example, they decrease breakdown voltage and increase leakage current. However, because primarily low-surface-energy planar defects grow, particularly during compound single crystal growth, they have been difficult to eliminate thus far.
Even in the case of homoepitaxially growing in which homogeneous crystals are grown, propagative planar defects propagate in the homoepitaxially grown layers of the substrate that is grown (for example, when the surface energy of the polar face exposed at the surface is low), making it difficult to obtain a homoepitaxially grown layer suitable as an underlayer in power semiconductor devices and the like.
For example, when silicon carbide (SiC) grows heteroepitaxially on an Si substrate, planar defects (anti-phase boundaries and twins) develop in on SiC film. One method of reducing such planar defects, comprising the step of providing a growth region on a grown substrate and a step of growing a single crystal silicon carbide crystal on the growth region to a thickness identical to or exceeding the natural thickness of the substrate along the orientation of the growth surface, has been proposed (Japanese Examined Patent Publication (KOKOKU) Heisei No.6-41400). However, the two types of reverse phase regions in the SiC have a characteristic of expanding in a mutually perpendicular orientation as the thickness of the SiC film increases. Thus, it is impossible to effectively reduce the anti-phase boundary no matter how long heteroepitaxial growth is continued.
Another method of eliminating the mismatching of crystal lattices is to insert a new layer (buffer layer) having a lattice constant falling between those of the substrate and the layer grown over it. However, for example, when growing SiC on a Si substrate, no substance capable of serving as buffer layer is known. As a result, when SiC is grown directly on a Si substrate, there is 20 percent of mismatching between lattice constants. This is the main cause of planar defects in epitaxially grown SiC layers.
In the growth of gallium nitride (GaN), to eliminate high-density crystal defects generated between heterogeneous substrates, a layer temporarily blocking GaN growth is provided part way through the growth process. It has been proposed to reduce crystal defects growing in a vertical direction by utilizing lateral growth of GaN above an opening provided in part of the growth-blocking layer, wherein the GaN has selectively been grown through the opening (U.S. Pat. No. 6,051,849). However, this method requires the steps of forming a GaN growth-blocking layer and making a hole in it, which increases cost. Further, since crystal defects have the property of propagating unchanged into the GaN over the opening, crystal defects are not completely eliminated.
Accordingly, the objects of the present invention are to provide a compound semiconductor crystal substrate for epitaxial growth capable of reducing planar defects such as twins and anti-phase boundary occurring in epitaxially grown crystals without additional steps beyond epitaxial growth, and to provide a method of manufacturing the same.
A further object of the present invention is to provide a laminate employing the above-described compound semiconductor crystal substrate and a method of manufacturing the same.